Mizuno reports increased excess heat

Quote from RobertBryant

Ascoli's next assertion may be that Mizuno has put in a random fudge factor.RDM

BlowerPwr = [RDM ] *(1.722298 * Airspeed - 3.484167)

I'm pre-empting that.

ascoli's assertion, as I understand it, is that the stated airspeed is not measured, but in fact calculated from measured blower power and the known (see paper) blower power / air speed relationship.

This is careless presentation of results; it should be made clear that the airspeed figure is so calculated, not measured. It does not affect the results. It very slightly, not greatly, affects the credibility of the paper (any carelessness makes readers more likely to expect other carelessness).

Quote from Grafiker

You'll see from my posting history that I usually make myself look foolish when trying to make a technical contribution, so I fully expect this to be shot down in flames... but here goes anyway!

Such questions as THH, Ascoli raise are OK, if we see only about 20-30% excess heat and have a low COP.

If you are used to cook in a kitchen then you very well know the difference between a lukewarm cooking plate and a glowing one. Errors there are really painful.

People that ask such question after the experiment shows a COP >8 are no longer serious, if they connect their reasoning with obvious errors in the experiment. E.g. Ascoli did as usual not even read Jed's paper. He just glanced at a single graphic. But what else should we expect? If such people do believe P&F did use beer for their experiments?

Quote from anonymous

There is NO RANDOM number added to the Mizuno numbers. (If this was Ascoli's hypothesis this is wrong.) It is the straight formula that I show:

Airspeed = 0.583436 * BlowerPwr + 2.010436

I never talked about random numbers added to any calculation. There is no random number. The whole set of the Blower Power (BP) and Air Speed (AS) data published by graph (1) and numbers (2) can be explained by two calculations and by the effect of multiple rounding off on primary (measured) and intermediate values:

- Blower Power was obtained mathematically by multiplying the Voltage and Current to the blower. Presumably, only the voltage was directly measured by an instrument. The current was probably obtained multiplying the voltage drop through a resistor by the inverse of its resistance;

- Air Speed was obtained mathematically by using a formula which fits the curve shown in Figure 4 of (3). This formula is probably similar to the exponential relationship shown in the formula (5) described at page 14 of a 2017 paper (4). However, due to the very narrow range of the 49 data sample provided by JR, the exponential relationship between BP and AS is undistinguishable from the linear one that you have proposed.

Quote

Also, the voltages and currents produce powers in Ascoli's table which don't match the tabular powers that Jed just provided us. The new table shows powers at 14 levels between 3.7010 and 3.7237 watts for the 49 data points, with a mean of 3.71205. Ascoli shows powers in the 4.5 to 4.7 watt range. Does Ascoli have some other data for this secondary hypothesis for rounding?

As you correctly noticed, the numbers provided by JR (2) include only 14 different values in a series of 49 pairs of data. This is true for both the BP and the AS series and, more importantly, the two values are always coupled in the same way up to the fifth significant figure. This only fact is by large more than sufficient to affirm with absolute certainty that the values of the two series are mathematically tied each other!

This is the list of all the 14 possible pairs of values:

The BP values can be subdivided into 9 main levels (highlighted in bold in the above table). Furthermore, 5 consecutive main levels have also a lower sublevel. The following table reports these levels, after having multiplied their values by 10000 in order to remove the decimals.

This sequence of quantized values (*) simply derives from the effects of multiple rounding off of measured data.

(*) I take the occasion for thanking THH for his careful analysis of this approach and for having explained it in a better way than I could ever do

The following is a possible set of V and I values whose product gives Power values very close to those listed in the first table:

The value of the resulting V and I data, on columns 3 and 4, comes from the two "V&I base values" incremented by the Delta V and Delta I listed in the first 2 columns.

The Delta V column lists 9 values rounded off at the second decimal digit.

More interesting is the second column which shows that the 5 sublevels in the power data come from a difference of -0.00003 A with respect to the base level. This unusual delta value shows that the rounding off affected a precursor of the current, probably the voltage across a resistor. It's very likely, that the digit "3" in the Delta I value is connected with the 3 ohm resistor mentioned by JR in a previous comment (5).

The last 3 columns show the values of the Power. The 3rd last column shows the complete numbers with all the decimal digits coming from the product of VxI. Due to the 2 decimal digits of the V values and the 5 decimal digits of the I values, the resulting Power numbers have 7 decimal digits, 3 more decimal digits than the Blower Power in the table published by JR (1), which is reported in the last column. The second last column shows the effect of the rounding off on the resulting BP values and shows how they are very close to the JR data.

(1) Mizuno reports increased excess heat

(2) Mizuno reports increased excess heat

(3) http://lenr-canr.org/acrobat/MizunoTincreasede.pdf

(4) https://www.lenr-canr.org/acrobat/MizunoTpreprintob.pdf

(5) Mizuno reports increased excess heat

Quote

GSVIT 's fluid is tight and holy water.

Just to clarify a couple of things I wrote, I have no problem with anyone taking a microscope to Mizuno's methodology. That sort of detailed and careful examination is never bad. It's just that to account for a power output of 250W with an input of 50W, any error to cause entirely wrong results for that is going to be a whopper-- not something subtle.

As to liquid flow calorimetry as opposed to air flow, the thermal properties of water (or other heat transfer liquids) are much better suited to making straightforward and accurate heat flux measurements in the required temperature and heat flow regime than is air. It is less likely heat or coolant will escape undetected through unknown pathways in a calorimeter constructed and implemented like SGVIT's than it is in Mizuno's. Still, as I have been beating the dead horse over, with this much power and this much power ratio, and those sterling calibration results, differences between air and water calorimetry almost certainty don't matter. So, apart from dishonesty or delusion, what we are probably looking to rule out is something very gross like the error JedRothwell cited NASA made when it confused unit systems causing a Mars probe to crash. Furthermore, it would have to be something that allows the calibration to come in correct but affects the active experiment result differently. That is a tall order.

I am still skeptical about the validity of the claims. However, I am at a complete loss to explain how an error meeting the above requirements could possibly be made under the circumstances. That makes for a very interesting situation indeed.

One other thing: I hope the always from the start idiotic idea that skeptics can't be convinced by good data has been defeated for good around here. Otherwise, name one skeptic on the forum who claims this Mizuno experiment couldn't be valid based only on a priori reasoning and concepts. Name one who has said they wouldn't accept the data if the experiment can be properly and reliably replicated by credible people or organizations. So please: stop all the dumb sh*t about pathological skepticicism already. Prior doubts have been logical, not pathological.

I never talked about random numbers added to any calculation. There is no random number. The whole set of the Blower Power (BP) and Air Speed (AS) data published by graph (1) and numbers (2) can be explained by two calculations and by the effect of multiple rounding off on primary (measured) and intermediate values:

Just to clarify, because the confusion is perhaps my fault.

Ascoli was looking at the effect of quantisation. Quantisation adds noise to a signal which is not random but has many of the characteristics of random numbers, and in analysis is often modelled as random. Specifically the quantisation noise is not linearly correlated to any physical quantity - which is why finding something else correlated with the quantisation noise shows that cannot be aphysical measurement.

I am still skeptical about the validity of the claims. However, I am at a complete loss to explain how an error meeting the above requirements could possibly be made under the circumstances. That makes for a very interesting situation indeed.

Trying to find errors of this sort is a mugs game. They are unknown unknowns and guesses tend to be wrong.

But to answer your challenge. The R20 result was a single sample. One option would be:

M infers I measurements from V and R, or indeed infers P from V and prior measurements

M, by mistake, forgets that the active and cal reactors have different heaters with completely different R. For example, he uses the old R19 cal reactor, but the R20 reactor has a different design of heater with a much lower resistance. Even, R20 heater short circuits and although designed normally has a much lower resistance in operation. If input power was calculated from supply voltage and previous measurements (as was done for the airspeed from the blower power) it could be very wrong.

In fact the same category error could mean that the wrong blower calibration curve was used to calculate airspeed for R19 measurements: though since those results are more substantive, and more carefully documented, that would be less likely.

This is a mistake that would be picked up in time: but maybe not for a single sample. Stranger things have happened, with no bad will. The R20 measurements are given without information on how input power is measured, so this is possible.

Quote from THHuxleynew

Just to clarify, because the confusion is perhaps my fault.

Ascoli was looking at the effect of quantisation. Quantisation adds noise to a signal which is not random but has many of the characteristics of random numbers, and in analysis is often modelled as random. Specifically the quantisation noise is not linearly correlated to any physical quantity - which is why finding something else correlated with the quantisation noise shows that cannot be aphysical measurement.

No, not your fault. The randomization argument was first used by JR in the vain effort to confirm his initial statement that the two curves in his graph came both from physical measurements (1). Then, he improperly attributed this possible interpretation to me (2).

On the contrary, you did an egregious work in explaining my analysis with the appropriate wording, especially by introducing the quantization analogy. Thanks, again.

You are also right in saying that data quantization, such as those due to truncation errors in the representation of numbers, could seem random errors. It happens when the length of the available data (ie the numbers of decimal digits) have been shortened with respect to the original values. This is exactly the situation described in my previous comment. The Power numbers provided by JR have only 4 decimal digits, but the original values coming from the VxI products had 7 decimal digits. This lack of significant figures in the available data introduces some irregularities in the difference between corresponding values, such as those shown in the second table of my previous comment. In that table, you can see that the differences between the main level are usually 25 points, but in two occasions they are 26 points. The same for the differences between two sublevels: it is usually 4 points, but in a case it is 5 points. These discrepancies, which give the impression of a random effect, are due instead to the fact that those differences were calculated starting from numbers rounded off to the 4th decimal digit.

(1) Mizuno reports increased excess heat

(2) Mizuno reports increased excess heat

Jed,

Perhaps something to ask Mizuno about the preliminary R20 measurement in particular. For the active reactor, how was the 50W input determined? Was the given result power (as we would hope) measured from real V and I measurements made contemporaneously through the experiment?

Real matters: for example inferring I from previously measured V/I curve of the reactor would be less safe.

THH

Mizuno's spreadsheet would give the same numbers shown here, if he had used that formula. Spreadsheets don't make rounding errors with this many digits. No formula with three terms will give the actual air speed numbers. You might come up with something if you keep piling on terms, but why would anyone do that? If Mizuno was trying to fool you, he would just use a random number generator. He would not go to the trouble to make a multi-term equation that hides the transformation. If he (or I) wanted to use the motor power instead of the anemometer reading, we would do that, and say that is what we did. We would not make an elaborate equation to do that. Multiplying by one constant (0.583) comes close enough for any practical purpose.

Jed: do you agree that the quantisation noise from the V and I values used to determine blower power is replicated in the airflow speed data? If not, what is your explanation for the very precise correlation between the V*I quantisation noise and the stated airspeed value as shown in ascoli's figure? No-one is saying Mizuno is trying to fool anyone - he is just using calculated rather than measured airspeed. The paper shows the relationship which presumably he is using.

I'll repeat ascoli's explanation for that correlation:

The whole set of the Blower Power (BP) and Air Speed (AS) data published by graph (1) and numbers (2) can be explained by two calculations and by the effect of multiple rounding off on primary (measured) and intermediate values:

- Blower Power was obtained mathematically by multiplying the Voltage and Current to the blower. Presumably, only the voltage was directly measured by an instrument. The current was probably obtained multiplying the voltage drop through a resistor by the inverse of its resistance;

- Air Speed was obtained mathematically by using a formula which fits the curve shown in Figure 4 of (3). This formula is probably similar to the exponential relationship shown in the formula (5) described at page 14 of a 2017 paper (4). However, due to the very narrow range of the 49 data sample provided by JR, the exponential relationship between BP and AS is undistinguishable from the linear one that you have proposed.

Obviously we don't know exactly how it was done: but we do know it was done because the V*I quantisation noise waveform correlating closely with the airspeed data proves this. That could not possibly be coincidental, and there is no physical mechanism for the measured quantisation on V*I to affect the measured airspeed.

Quote from JedRothwell

Yes, it is one sample. But there are 55 other samples shown in Table 1, and hundreds of other samples and calibrations from other data, and -- here is the point: It was the same calorimeter. It works the same way every time. If there was a problem, it would show up in all the other measurements of other samples. So why are you pretending the R20 result was sui generis? The other tests produced 40 to 100 W, which is in the same range as 250 W. If there was a problem with 250 W, it would also be seen at 40 to 100 W.

If you find a significant error in the graphs for R20, I will find that same error in the calibrations for 10 W, 50 W, 100 W, and the excess heat runs in Table 1. It will be readily apparent. So you can use this data to search for a problem. But you are saying there is one run from that reactor, at that power level, in this report, so you cannot draw conclusions. As if it were somehow unique. That's incorrect. The calorimeter works the same way not matter what reactor you put in it. I have a table summarizing calibration data from 4 different reactors, of different sizes. The calorimeter response is identical. You cannot tell which one is in it. (The reactor surface temperatures are somewhat different at the same power level, because they have different surface areas.)

That is a good point Jed. The difference between R20 and R19 was the active reactor heater element. I have suggested above a mechanism which (via a mistake) could underestimate the R20 active reactor input power by any (possibly varying) amount and not affect R19, nor the calibration data. If we had a large quantity of active reactor R20 data, at different powers, we could test this. More specifically, if we have the raw data for the R20 active runs, with V and I in to heater, we could easily rule this out or confirm it.

Excellent Jed, so if we had the R20 raw data we could check this and confirm that Mizuno used measured V and I on heater to get 50W, rather than an assumption that the same voltage would generate the same power and a previously determined voltage / power relationship?

This is no different than using calculated airspeed (from blower power and characteristics). It is not unreasonable, just a bit less safe.

Why would the R20 be different?

(1) it is a sample measurement, not part of a systematic set

(2) the heater element has different geometry and possibly requires different drive

(3) he might not have had access to suitable heater current measurement, or have been using that for something else at the time.

(4) the results were collected at a different time. Possibly Mizuno felt confident enough that he could rely on just a voltage measurement?

Perhaps I could ask you, where does Mizuno say it is the same? There is no requirement for different experiments to have exactly the same data collection and analysis methods.

We have hundreds of calibrations from other reactors. Over 100 are summarized in Figs. 1, 2 and 3. We don't need the R20 data. If you think there is something unique about the R20 data that makes the previous calibrations not applicable, I suggest you ignore it and look only at the Table 1 results.

120W calibration raw data: 67.67V 1.77A => 38.2 ohms

Stated M2 heater: 100V 500W => R = 10,000 / 500 = 20 ohms

The R20 heater element would appear to be different from the earlier (120W calibration) element.

Jed, I am asking for this for the R20 result because of what Mizuno says. It has a different heater geometry, it is a different reactor, the result is very much larger, there is only one result. The above rough calculation appears to show it is a different heater element from that used in 2017 tests.

If Mizuno assumed everything was the same as in previous reactors he might use voltage measurement only and the known voltage power relationship. Just as he used measured power and known power airspeed relationship for the airspeed values. That could result in an error, given this is only one result and maybe not so carefully checked.

Obviously, because this result is very significant, if the raw data existed and this possibility could be ruled out that would be useful. It would still only be one result, but a better validated one.

Last Mizuno patent application JP246650215 06.06.2019

Quote from JedRothwell

I have it, and I confirm it. Why would I tell you that if I did not have this, and dozens of other spreadsheets from the past 10 years? Look at the spreadsheets I uploaded previously. They all show amps and volts recorded every 5 seconds.

What do you mean "assumption that the same voltage would generate the same power"? Why would he not measure amperage? Who wouldn't measure that??? Does it say anywhere, in any report of schematic that he only measures voltage and not amperage? Do you think he assumes a constant amperage power supply always works? Or that he wouldn't set up the data collection to work just a well with a constant voltage setting? In fact, they both change in when input power increases in Fig. 6.

Have you ever heard of an experimentalist who does not measure both amperage and voltage? That is such a weird thing to say!

Here is some sample data from the R20 spreadsheet shown in Fig. 6. This data is Minute 1 (38 W) and Minute 20 (50 W). Multiply the raw data numbers by 60 to compute watts.

V	A	Watts (V*A*60)
Data from 0:01
1.1730	0.5400	38.0052
1.1730	0.5400	38.0052
1.1730	0.5400	38.0052
1.1730	0.5400	38.0052
1.1730	0.5400	38.0052
1.1785	0.5425	38.3602
1.1785	0.5425	38.3602
Data from 0:20
1.3370	0.6230	49.9771
1.3370	0.6230	49.9771
1.3370	0.6230	49.9771
1.3370	0.6230	49.9771
1.3370	0.6230	49.9771
1.3375	0.6230	49.9958
1.3370	0.6230	49.9771

Display More

Thanks Jed, that is very helpful.

I don't understand the X60 adjustment? Could you possibly explain it?

My understanding is that P = V * I.

The previous raw data did in fact show P = V * I so there appears to be some difference in the measurement setup here.